Real-time risk monitoring of ship pilotage operations: Automating BN risk model development

The maritime industry is evolving further with new technologies and services, such as autonomous ships and remote pilotage operations, under development. Whilst these services may bring new opportunities and benefits, the ability to manage risks becomes increasingly vital. Furthermore, these new technologies and services require risk models of a dynamic nature, which can incorporate ongoing systemic changes and provide risk estimation in real time. Hence, this paper presents a novel approach, which extracts the incident data and automatically establishes a real-time Bayesian risk model. The model consists of a clear hierarchy denoting a chain of risk events, i.e., root causes, hazards, accidents, and losses. The resulting risk model provides an estimation of the posterior probability of occurrence of all variables in the Bayesian network. These results are next plotted and monitored with Graphical user interface application to monitor the critical factors leading to losses and thus requiring risk control in real time. The effectiveness of the method is afterward demonstrated using a case study of ship pilotage operations. The resulted model shows the probability of occurrence of risk events such as accidents, losses, hazardous scenarios and causal factors.